Naltrexone has been shown to reduce the risk of relapse to heavy drinking in alcoholics. Its approval by the FDA for the treatment of alcohol dependence makes elucidation of the range of its potential applications of considerable scientific and clinical importance. Recent naltrexone trials in alcoholics have shown that compliance with the medication is variable and that it is superior to placebo only among alcoholics who are highly compliant with the medication regimen. The availability of a long-acting depot formulation (LADF) of naltrexone, which in preliminary studies has been shown to be safe and well tolerated, provides a potential means to enhance compliance with the medication. In this study, we propose to conduct a three-arm, parallel-groups comparison of a LADF of naltrexone, daily oral naltrexone, and placebo in a sample of 150 DSM-IV alcohol-dependent subjects. Random assignment to treatment group and strict separation of clinical and research activities will be maintained throughout the study. A double-dummy procedure, in which placebo forms of both the depot and oral formulations will be used, will serve to maintain the double-blind design. Subjects will receive treatment for a total of three months, with two subsequent posttreatment follow-up visits spanning a six-month period. There are two principal hypotheses of the study: 1) the LADF of naltrexone will be comparable in safety. but significantly more efficacious than a placebo injection, in the prevention or moderation of relapse to heavy drinking and 2) that the long-acting depot and oral formulations of naltrexone will be comparable in safety, but in an intention to treat analysis, the LADF will be significantly more efficacious in the prevention or moderation of relapse to heavy drinking. Secondary hypotheses are that 1) the effects of the LADF will persist longer than those of oral naltrexone, based upon greater improvement in the LADF naltrexone group during active treatment and 2) that it will be possible to identify valid predictors of compliance based on pretreatment demographic variables, severity of alcohol dependence and a measure of "readiness to change" drinking behavior. Information obtained from this study may be expected to increase pharmacotherapeutic options in alcohol- dependent patients and clarify which patient predictors can be used to optimize therapy.